Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Pupil dilation during effortful listening reflects individual differences in context sensitivity
Poster Session E, Sunday, September 14, 11:00 am - 12:30 pm, Field House
Naomi Sellers1, Hannah Mechtenberg1, Emily Myers1; 1University of Connecticut
Introduction: Successful speech comprehension requires the appropriate allocation of cognitive resources, reflecting the amount of “listening effort” an individual employs to reach their listening goal. The amount of effort needed to perform well on a given listening task varies depending on a number of factors including the listening environment, the listening task at hand, and the differences in the listener themself. One factor that may influence listening effort is contextual support from the semantic predictability of the spoken materials. Sentence-based semantic predictability has been shown to modulate neural responses, increases processing speed and accuracy after speech perception, and even improves comprehension of degraded speech signals. These studies tend to find that when the number of words likely to complete a sentence is restricted by the sentence context (i.e., there is high sentential constraint), and when the word completing the sentence is one of those options with high likelihood, neural responses are smaller. Moreover, newer work suggests that the processing benefit from predictive contexts might vary across individuals. Yet, how semantic predictability leads to these patterns of improvement is still uncertain, and semantic predictability’s impact on listening effort is still understudied, especially at the individual level. This study used pupillometry to examine how semantic predictability affected the allocation of listening effort, and how this effort modulation systematically varied across participants. Pupillometry is a well-supported measure of listening effort and has been associated with the phasic norepinephrine activity in the locus coeruleus. Methods: Pupil dilation was recorded while young adult participants (n = 64) with normal hearing listened to sentences varying in semantic predictability. Sentence stimuli were mixed with background noise at a challenging signal-to-noise ratio, and participants completed a probe-word verification task after hearing each sentence. Pupil results were modeled using growth curve analysis to assess the dynamics of semantic predictability during online speech perception. Predictors of individual variation in predictive context sensitivity were derived from sentence predictability’s influence on probe-word response and from deviation from a separate, secondary cloze task. Results: Participants were indeed more accurate at identifying probe words embedded in highly predictable than less predictable sentences; importantly, this accuracy benefit for semantic predictability varied across participants. While sentence predictability did not significantly influence group-level pupil dilation directly, a post-hoc analysis showed that the magnitude of pupil dilation correlated with the individual degree of benefit from predictive semantic contexts. Conclusion: These data suggest that listeners differentially use contextual information to guide effortful processing during speech perception, leading to new insights into individual-level engagement with the speech signal. Future work is needed to explore the contributors to these observed differences, such as how prior language experience affects type and strength of input predictions, variation in working memory capacity and attentional control, and if there is a neurobiologically-based limiting capacity of cognitive resources like firing rate of the locus coeruleus or differences in sensitivity to norepinephrine in the arousal system.
Topic Areas: Speech Perception, Meaning: Lexical Semantics